Loading...

Change Owns to Parent Ofs

Set start and due date...

Xporter

XML

Word

Printable

Details

Type: Feature
Priority: Should have
Fix Version/s: PI19
Component/s: COM PSS SW
Labels:
None

ARTs:

Data Processing
Benefit hypothesis:

Hide

Provides rigid, reproducible measurements of the performance of different implementations of the PSS fast transient search pipeline and helps towards converging on an architecture for it based on those measurements.

Show
Provides rigid, reproducible measurements of the performance of different implementations of the PSS fast transient search pipeline and helps towards converging on an architecture for it based on those measurements.
Acceptance criteria:
Hide

Documentation that

describes the set up of the benchmarking tests and instructions for running the tests

summarises the results and identifies the bottlenecks

describes the outlook based on the results, focusing on current SPS performance against real time
Show
Documentation that describes the set up of the benchmarking tests and instructions for running the tests summarises the results and identifies the bottlenecks describes the outlook based on the results, focusing on current SPS performance against real time
Feature Points:
1
Initial Size:
1
WSJF:
0
Epic Link:
Data Processing Pipeline Development for AA2 onwards
Agile Teams:

Team_PSS
Due Sprint:
Sprint 5
Story Point Burn-up:
Overdue:
Outcomes:

Hide

This work has been written up in this document:
https://www.overleaf.com/read/vztqwgntrpvc

The benchmarking results are different from the standalone tests. The SPS pipeline integrated into the CHEETAH is very efficient and can be run on four threads for a MID beam. However, all the processes are running in series CT+DDTR+SPDT. We can significantly improve the performance but working on changing the cheetah framework to accommodate the Muti-threading functionality inside a task.

The typical (execution time/ realtime) performance measure is ~0.7. We must do further efficacy and long-term stability tests in the following PIs.

Show
This work has been written up in this document: https://www.overleaf.com/read/vztqwgntrpvc The benchmarking results are different from the standalone tests. The SPS pipeline integrated into the CHEETAH is very efficient and can be run on four threads for a MID beam. However, all the processes are running in series CT+DDTR+SPDT. We can significantly improve the performance but working on changing the cheetah framework to accommodate the Muti-threading functionality inside a task. The typical (execution time/ realtime) performance measure is ~0.7. We must do further efficacy and long-term stability tests in the following PIs.
Resolved PI.Sprint:
19.6

Feature Checklist:

Stories Completed, Outcomes Reviewed, Satisfies Acceptance Criteria, Accepted by FO

Demos:
- DP_ART_20.1
Requirement Status:

PI22 - UNCOVERED
Goals_MIRO:
PSS-G1

Description

A reference CPU(AMD)-based Single Pulse Search pipeline is configured, with a data receptor, dedispersion transform and single pulse detector as the testbed for benchmarking the performance of its components and determining bottlenecks in that performance.

Additionally, functionality comparisons are made and documented between the CPU-based SPS and 'astroaccelerate' versions of SPS.

Attachments

Issue Links

mentioned in: Page Loading...

Structure

Activity

People

Assignee:: Noutsos, Aristeidis

Reporter:: Levin-Preston, Lina

Votes:: 0 Vote for this issue

Watchers:: 0 Start watching this issue

Feature Progress

Story Point Burn-up: (100.00%)

Feature Estimate: 1.0

	Issues	Story Points
To Do	0	0.0
In Progress	0	0.0
Complete	4	10.0
Total	4	10.0

Dates

Created:: 10/May/23 1:26 PM

Updated:: 15/Apr/24 9:37 AM

Resolved:: 07/Sep/23 1:46 PM

Configure a CPU-based Single Pulse Search pipeline for benchmarking with quasi real-time data input